symm/salsa20-arm-neon.S: Mark the final-permutation stores as word-aligned.

This was just an oversight when I was hacking the initial code.

diff --git a/symm/salsa20-arm-neon.S b/symm/salsa20-arm-neon.S
index 15585e4..cea942d 100644 (file)
--- a/symm/salsa20-arm-neon.S
+++ b/symm/salsa20-arm-neon.S
@@ -213,25 +213,25 @@ FUNC(salsa20_core_arm_neon)
        vadd.u32 q10, q10, q14
        vadd.u32 q11, q11, q15
 
-       vst1.32 {d16[0]}, [r2]!
-       vst1.32 {d22[1]}, [r2]!
-       vst1.32 {d21[0]}, [r2]!
-       vst1.32 {d19[1]}, [r2]!
-
-       vst1.32 {d18[0]}, [r2]!
-       vst1.32 {d16[1]}, [r2]!
-       vst1.32 {d23[0]}, [r2]!
-       vst1.32 {d21[1]}, [r2]!
-
-       vst1.32 {d20[0]}, [r2]!
-       vst1.32 {d18[1]}, [r2]!
-       vst1.32 {d17[0]}, [r2]!
-       vst1.32 {d23[1]}, [r2]!
-
-       vst1.32 {d22[0]}, [r2]!
-       vst1.32 {d20[1]}, [r2]!
-       vst1.32 {d19[0]}, [r2]!
-       vst1.32 {d17[1]}, [r2]!
+       vst1.32 {d16[0]}, [r2 :32]!
+       vst1.32 {d22[1]}, [r2 :32]!
+       vst1.32 {d21[0]}, [r2 :32]!
+       vst1.32 {d19[1]}, [r2 :32]!
+
+       vst1.32 {d18[0]}, [r2 :32]!
+       vst1.32 {d16[1]}, [r2 :32]!
+       vst1.32 {d23[0]}, [r2 :32]!
+       vst1.32 {d21[1]}, [r2 :32]!
+
+       vst1.32 {d20[0]}, [r2 :32]!
+       vst1.32 {d18[1]}, [r2 :32]!
+       vst1.32 {d17[0]}, [r2 :32]!
+       vst1.32 {d23[1]}, [r2 :32]!
+
+       vst1.32 {d22[0]}, [r2 :32]!
+       vst1.32 {d20[1]}, [r2 :32]!
+       vst1.32 {d19[0]}, [r2 :32]!
+       vst1.32 {d17[1]}, [r2 :32]!
 
        // And with that, we're done.
        bx      r14